A new feature like this should definitely start as a community extension, this gives more room to play around how the feature should work, and lets us see how much interest there is. I see the existing extension will soon reach 1000 downloads, that's a good start, but there are many community extensions much more popular than that.
The other aspect to consider is whether the extension would benefit from being tied more closely to the Flarum API. This need would arise if we start seeing multiple other extensions depending on it, and it would probably make sense to create an official API for it, so third-party extensions can integrate with the feature more reliably. For an extension that just extends or replaces the login, we've seen many community extensions do this reliably already, so there's probably no change in Flarum Core required.
I see there were some suggestions in the discussion for the existing extension. The logical next step would probably be to try it out, and confirm/update the list of missing features. Then a bounty could go towards perfecting the existing extension, forking it, or creating a new extension with a detailed list of differences from the existing extension.
There's no reason to hold back from using an existing extension. If it works well, you can save up your bounty money to sponsor an update for Flarum 2.0 when the time comes. Apart from that, it's unlikely much maintenance would be necessary. Unless it depends on still-evolving web standards? I haven't checked.
